Large Scale Parsing of Czech
Pavel Smrž, Aleus Horák · International Conference on Computational Linguistics · 2000
Syntactical analysis of free word order languages poses a big challenge for natural language parsing. In this paper, we describe our approach to feature agreement fulfilment that uses an automatically expanded grammar. We display the insides of the implemented system with its three consecutively produced phases --- the core meta-grammar, a generated grammar and an expanded grammar. We present a comparison of parsing with those grammar forms in terms of the parser running time and the number of resulting edges in the chart, and show the need of a shared bank of testing grammars for general parser evaluation.